carbonaceous pm2.5 and secondary organic aerosol across
TRANSCRIPT
1
Author’s postprint version of the article on institutional repository after 12
months embargo from first online publication
(Published online: 15 January 2016)
Carbonaceous PM2.5 and secondary organic aerosol across the
Veneto region (NE Italy)
Md. Badiuzzaman Khan
a, Mauro Masiol
b,
Gianni Formentonc, Alessia Di Gilio
d,e,
Gianluigi de Gennarod,e
, Claudio Agostinellia,
Bruno Pavonia*
Published in
Science of The Total Environment
Volume 542, Part A, Pages 172-181
https://doi.org/10.1016/j.scitotenv.2015.10.103
https://www.sciencedirect.com/science/article/pii/S0048969715309190
Released under a Creative Commons Attribution Non-Commercial No Derivatives
License
2
CARBONACEOUS PM2.5 AND SECONDARY
ORGANIC AEROSOL ACROSS THE VENETO
REGION (NE ITALY)
Md. Badiuzzaman Khana, Mauro Masiol
b,
Gianni Formentonc, Alessia Di Gilio
d,e,
Gianluigi de Gennarod,e
, Claudio Agostinellia,
Bruno Pavonia*
aDipartimento di Scienze Ambientali
Informatica e Statistica, Università Ca’ Foscari Venezia
Dorsoduro 2137, 30123 Venezia, Italy
bDivision of Environmental Health and Risk Management
School of Geography, Earth and Environmental Sciences
University of Birmingham
Edgbaston, Birmingham B15 2TT
United Kingdom
cDipartimento Provinciale di Padova
Agenzia Regionale per la Prevenzione e Protezione Ambientale del Veneto
(ARPAV), Via Ospedale 22, 35121 Padova, Italy
dAgenzia Regionale per la Prevenzione e Protezione Ambientale della Puglia
(ARPAP), Corso Trieste 27, 70126 Bari, Italy
eDipartimento di Chimica, Università degli Studi di Bari, Via Orabona 4, 70126,
Italy
(*) Corresponding Author: [email protected]
3
ABSTRACT
Organic and elemental carbon (OC-EC) were measured in 360 PM2.5 samples collected from April
2012 to February 2013 at six provinces in the Veneto region, to determine the factors affecting the
carbonaceous aerosol variations. The 60 daily samples have been collected simultaneously in all
sites during 10 consecutive days for 6 months (April, June, August, October, December and
February). OC ranged from 0.98 to 22.34 µg/m3, while the mean value was 5.5 µg/m
3, contributing
79% of total carbon. EC concentrations fluctuated from 0.19 to 11.90 µg/m3
with an annual mean
value of 1.31 µg/m3
(19% of the total carbon). The monthly OC concentration gradually increased
from April to December. The EC did not vary in accordance with OC. However the highest values
for both parameters were recorded in the cold period. The mean OC/EC ratio is 4.54, which is
higher than the values observed in most of the other European cities. The secondary organic carbon
(SOC) contributed for 69% of the total OC and this was confirmed by both the approaches OC/EC
minimum ratio and regression The results show that OC, EC and SOC exhibited higher
concentration during winter months in all measurement sites, suggesting that the stable atmosphere
and lower mixing play important role for the accumulation of air pollutant and hasten the
condensation or adsorption of volatile organic compounds over the Veneto region. Significant
meteorological factors controlling OC and EC were investigated by fitting linear models and using a
robust procedure based on weighted likelihood, suggesting that low wind speed and temperature
favour accumulation of emissions from local sources. Conditional probability function and
conditional bivariate probability function plots indicate that both biomass burning and vehicular
traffic are probably the main local sources for carbonaceous particulate matter emissions in two
selected cities.
Keywords: PM2.5, Organic carbon, Elemental carbon, Meteorological factors, Secondary organic
aerosol, Long-range transport, Po Valley
4
1. INTRODUCTION
Carbonaceous compounds account for an important fraction of atmospheric particulate matter (PM),
generally contributing between 30 and 50% of PM with aerodynamic diameter < 2.5 μm (PM2.5) in
many European urban areas (Putaud et al., 2004). In the recent years, the scientific community has
paid much attention to the carbonaceous fraction of PM, since it may affect the global radiation
budget, cloud microphysics (Seinfeld and Pandis, 1998; Lyamani et al., 2006) and has effects on the
global climate change (Hitzenberger et al., 1999; Dan et al., 2004). In addition, increased levels of
urban carbonaceous PM are significantly associated with cardiovascular mortality and morbidity
(Na et al., 2004; Ito et al., 2011) and several organic compounds of known or suspected
carcinogenic and/or mutagenic potential are commonly found in PM, e.g., polycyclic aromatic
hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) (WHO, 2000).
The carbonaceous PM is composed of a refractory component, commonly called elemental carbon
(EC) and an organic fraction (OC). EC is a primary pollutant being mainly released in particle-
phase from the incomplete combustion of carbon-containing fuels. On the contrary, OC includes
thousands of different organic compounds (e.g., aliphatic, aromatic hydrocarbons, carboxylic acids
and carboxylic compounds with polar substituent, etc.) with widely varying chemical and physical
properties, such as molecular weights, level of oxidations and gas-particulate partitioning. OC can
be directly emitted as a primary OC from many sources including combustions, industrial
emissions, geological and natural sources. It can also form in the atmosphere as secondary OC,
when some volatile and semi-volatile organic compounds (VOC and SVOCS) are chemically
transformed and the products undergo condensation or nucleation.
Although many recent researches have studied the carbonaceous fraction of PM, its chemical
characterization is not comprehensively accomplished yet and some formation mechanisms are still
unclear (Jacobson et al., 2000). In addition, the quantification of primary and secondary OC is a
5
very challenging task due to several reasons, such as the breaking down process of OC into primary
and secondary organic carbon, potential sampling and analytical artefacts, effects of meteorological
conditions during the sampling, etc. However, an estimation is possible through several indirect
methods. Among others, the EC-tracer method is a widely accepted technique (Turpin and
Huntzicker, 1995; Castro et al., 1999, Lim and Turpin 2002; Cabada et al., 2004; Harrison and Yin,
2008) and uses EC as a tracer of primary organic carbon (POC).
Several studies have been conducted across Italy on the carbonaceous PM (e.g., Lonati et al., 2005,
2007: Milan; Perrone et al., 2011: Milan; Larsen et al., 2012: Po Plain Italy; Cuccia et al., 2013:
Milan; Malaguti et al. 2013: Trisaia ENEA Research Centre) and a recent review of available data
was also recently published (Sandrini et al., 2014). However, data available for the Veneto Region,
NE Italy, are quite limited so far. This is a serious gap, as Veneto is one of the most industrialized
areas of Italy and is located in the Eastern part of the Po Valley, which is a well-known European
hot-spot for many air pollutants, including PM10, PM2.5 and PM10-bound PAHs (Masiol et al.,
2013). The presence of high mountain ranges, i.e. the Alps to the North and NW and the Apennines
to the South, protects the Po Valley from cold winds coming from the NE (Larsen et al., 2012). In
addition, the occurrence of frequent and prolonged wind calm periods, atmospheric stability
conditions, persistent thermal inversions and lower planetary boundary layer heights during the
night-time may favour the accumulation of locally-emitted pollutants especially during the coldest
months (Masiol et al., 2014).
This study aims to (1) quantify the seasonal and spatial trends of carbonaceous aerosols in the
Veneto Region; (2) investigate the effects of some meteorological factors on the carbonaceous PM
variations, (3) estimate the contribution of secondary organic carbon (SOC) formation to the
frequent exceeding of air quality standards for PM2.5 (4) investigate the potential local sources of
carbonaceous PM2.5 in two cities selected as proxies for all the cities in the region because of their
6
differences in emission scenarios and geographic features, and (5) account for the potential effects
of long-range transports on OC and EC levels.
2. MATERIALS AND METHODS
2.1. Study area
The Veneto Region has an area of ~18.4∙103 km
2, extending for 210 km in the north-south and 195
km in the east -west directions. This region is characterized by a mountainous area (29%) at north,
an intermediate hilly zone (15%) and a wide flat area located in the southern part. The
morphological characteristics and climatic conditions have a decisive effect on the distribution of
the population, which is mainly concentrated in the southern flat areas of the region, where there are
several major cities and a continuum of minor urban centres. There are seven administrative
provinces in Veneto: Belluno, Treviso, Vicenza, Venice, Padua, Verona and Rovigo.
2.2 Sample collection and analysis
A year-long sampling campaign (2012-2013) was conducted at 6 major cities located in 6 Provinces
(Figure 1). Five cities (named Belluno (BL), Vicenza (VI), Venezia-Mestre (VE), Padua (PD) and
Rovigo (RO)) are Province capitals, while the city of Conegliano (TV) was chosen for the Province
of Treviso. Unfortunately, the Province of Verona was not included in this study. All sampling sites
are operated by the Regional Environmental Protection Agency (ARPAV) and were chosen as
representative of city-wide emissions. The characteristics of each station are given in Table 1.
The PM2.5 samples were collected using low-volume samplers set according to the European
standard EN 14907:2005, i.e. with a nominal flow of 2.3 m3 h
-1. Samples were collected on quartz
fibre filters (47 mm Ø, Whatman QMA, GE Healthcare, USA) over 24 hours starting at midnight.
As reported in EN 14907:2005, filters were conditioned for a period of at least 48 h before and after
being weighted in a conditioning chamber equipped with a control system for constant temperature
7
and humidity (20±1 ºC and of 50±5% relative humidity) (Emerson S05KA Emerson Network
Power – Piove di Sacco-Pd). An analytical balance (Sartorius series Genius, mod. SE2, Germany)
with a sensitivity of 0.0001 mg was used to weight the PM mass (average of two measures).
Samples were stored in clean Petri slides at a temperature of -20 ºC to avoid sample degradation
and losses of the more volatile compounds.
Sixty samples per site were selected for subsequent chemical analysis in every alternate month
(April, June, August, October, December 2012 and February 2013): 10 samples per site in 10
consecutive days of the months selected. OC and EC concentrations were then determined on the
selected filters by using a Sunset Lab OC-EC Aerosol Analyzer (Sunset Laboratory Inc., USA) and
applying the NIOSH-5040 method. In brief, the speciation of OC and EC was achieved by placing a
punch (1 to 1.5 cm2) of filter inside a quartz oven, where the sample is heated in four increasing
temperature (ramps from 0 °C to 870 °C) in a completely oxygen free helium atmosphere in the first
phase, and in a 2% oxygen/helium mixture atmosphere in the second phase. He–Ne laser light
passing through the filter allows a continuous monitoring of filter transmittance, and thus the
correction for the pyrolytically generated OC. Carbon evolved during both phases, is measured by a
flame ionization detector (FID) after conversion to CH4 inside methanator. Instrument calibration
was achieved through injection of a known volume of methane into the sample oven for each
analysis (Birch and Cary, 1996).
2.3 Quality control
Field blank samples were also collected in order to evaluate possible contaminations on OC
concentrations. The calibration of the instrument was achieved by injecting on filter blanks four
standards of increasing concentration of sucrose (Sigma Aldrich, purity≥99%, ACS reagent). The
limit of detection (LOD = 0.15 μg-C cm-2
) of OC was calculated as three times the standard
8
deviation of fields blanks and the coefficient of variation (<5%) was calculated by performing ten
sample replicates for each calibration level.
2.4 Conditional probability function (CPF) and conditional bivariate probability function
(CBPF)
Conditional probability function (CPF) is an important tool for the identification of local point
sources (Kim et al., 2003; Kim et al., 2005) and it estimates the probability that a source is located
within a particular wind direction sector, ∆θ (Ashbaugh et al., 1985; Begum et al., 2010). CPF is
defined as:
𝐶𝑃𝐹∆𝜃 = 𝑚∆𝜃|𝑐>𝑥
𝑛∆𝜃
Where 𝑚∆𝜃 is the number of samples from the wind direction θ (22.5 degree in this study) having
concentration C higher than or equal to a threshold value x and n∆θ is the total number of samples of
the same wind sector ∆θ. Here, x represents a high percentile concentration (75th
or 90th
).
Recently a new technique was developed by Uria-Tellaetxe and Carslaw (2014) combining
conditional probability function with bivariate polar plots to detect and characterize the source of
pollutants. Bivariate polar plots show how a concentration of a species varies jointly with wind
speed and wind direction in polar coordinates and give directional information on sources as well as
wind speed dependence of concentrations. The conditional bivariate probability function (CBPF)
includes wind speed with CPF and allocates the observed pollutant concentration to a cell defined
by ranges of wind direction and wind speed. It is defined as:
𝐶𝐵𝑃𝐹∆𝜃,∆𝑢 = 𝑚
∆𝜃,∆𝑢 ⃒𝐶>𝑥
𝑛∆𝜃,∆𝑢
9
Where m∆θ,∆u is the number of samples from the wind direction ∆θ with wind speed interval ∆u
having concentration C higher than or equal to a threshold value x and n∆θ,∆u is the total number of
samples in that wind direction-speed interval. Details of the CPF and CBPF methods can be found
elsewhere (Uria-Tellaetxe and Carslaw, 2014). To estimate CPF and CBPF, daily measured OC, EC
values were allocated to each hour of a given day to match hourly wind speed and wind direction
data.
2.5 Back-trajectory analysis
A cluster analysis on the back-trajectories was performed to investigate the potential effects of long-
range transports on the levels of OC, EC and TC. Back-trajectories were computed by using the
HYSPLIT model (Draxler and Rolph, 2013; Rolph, 2013). The setup was: 96 h backward; starting
height at 20 m a.s.l.; 4 trajectories per day at 3, 9, 15 and 21 UTC calculated separately for all the
sites. A clustering algorithm using the Euclidean distance measure (Carslaw, 2014) was further
applied to group back-trajectories into clusters depending on their potential origin.
3. RESULTS AND DISCUSSIONS
3.1 Overview on results
Summary statistics of the atmospheric concentrations of PM2.5, Total Carbon (TC) and Total
Carbonaceous Aerosols (TCA) at the stations representing the six provinces of Veneto region are
listed in Table 2. PM2.5 concentrations varied from 3 to 83 µg m-3
with a mean [mean±standard
deviation (SD)] of 24±17 µg m-3
. The concentration of PM2.5 showed an increasing trend from
warmer months (June, 12 µg m-3
, August, 17 µg m-3
) to colder months (December, 41 µg m-3
,
February, 38 µg m-3
). The highest mean annual concentration was found in Padova (29 µg m-3
),
followed by Vicenza (28 µg m-3
), Rovigo (27 µg m-3
) and Venice (25 µg m-3
), whereas
comparatively lower mean concentrations were observed in Belluno (17 µg m-3
) and Treviso (18 µg
m-3
). These PM levels are comparable to the concentrations normally observed in Milan (33 µg m-
10
3, Lonati et al., 2005), (58 µg m-3
, winter, and 25 µg m-3
, summer, Lonati et al., 2007) and
Florence (annual mean value 19±10 µg m-3
, Giannoni et al., 2012). Atmospheric dispersion due to
increased wind speeds and wider mixing layer heights are recognised to be responsible for the lower
PM concentrations in warmer months (Ferrero et al., 2010; Pecorari et al., 2013): while the typical
planetary boundary layer height in the Po Valley is 450 m in winter (Bigi et al., 2012), it increases
up to 1500-2000 m during summer due to thermal convective activity (Di Giuseppe et al., 2012).
The total organic mass associated with OC (here designated as organic matter, OM) was calculated
by multiplying the measured OC by a factor of 1.4, as suggested by Lonati et al. (2005). TCA is
calculated as sum of OM and EC. The average annual contribution of TCA to PM2.5 is 37% ranging
from 18% to 84%.
The mean OC concentration was 5.5±4 µg m-3
(mean±SD) accounting for 22% of the total PM2.5
and 79% of the TC. On the contrary, EC varied from 0.2 to 11.9 µg m-3
with a mean value of 1.3 µg
m-3
and contributed to 5% of the total PM2.5 and 21% of the TC. A similar value was found by
Lonati et al. (2007) in Milan, where the OC value ranged from 0.4 to 35.8 µg m-3
with a mean
annual value of 9±7 µg m-3
and EC value ranged from 0.4 to 5 µg m-3
with mean annual value of
1.3 µg m-3
. The concentrations of OC and EC reported in various European cities are given in Table
3.
3.2 OC/EC Ratio
The OC/EC ratios ranged from 0.7 to 15.4 with an annual mean value of 4.5 (Table 4). A similar
result was observed in Milan (annual mean value 6.5) by Lonati et al. (2007) and at urban
background (from 2.1 to 8.4) and urban traffic (from 2.9 to 8.3) sites in Tuscany (Giannoni et al.,
2012). These values, which are similar to the values observed in most of the urban background sites
in the Po Valley (summer: from 5 to 7; winter: from 3 to 10) and also comparable to the rural sites
11
of the Po Valley, suggest a regional influence of SOA coming from anthropogenic and biogenic
sources in summer and the extensive wood burning utilization for residential heating in winter
(Larsen et al., 2012). Condensations of gas phase semi-volatile organic species due to lower
temperature combined with frequent events of air stagnation are responsible for increased OC/EC
ratios in the Po Valley (Gilardoni et al., 2011; Larsen et al., 2012; Sandrini et al., 2014). However,
the mean OC/EC ratios in this study were generally found above the values observed in most
European cities. The OC/EC ratios in summer and winter in Amsterdam, Barcelona, and Ghent
were 2.8-4.7, 2.6-3.1 and 3.5-4.4, respectively. The higher OC/EC ratios were referred to the
formation of secondary organic carbon in addition to primary organic carbon (Kim et al., 2000; Na
et al., 2004).
3.3 Spatial and seasonal trends
Monthly variations of OC and EC concentrations are shown in Figure 2. Generally the higher
concentrations were found during colder months, while lower concentrations were found in the
warmer ones. The highest OC level was observed in December (11.4 µg m-3
), the lowest in April
(2.2 µg m-3
). The monthly organic carbon concentration gradually increased from April to
December. Both the OC and EC concentrations were higher in February (9.1 µg m-3
), than in other
months except December. Although the monthly mean of EC concentration does not exhibit larger
variability, it follows the same pattern as OC with maxima during the colder months and minima in
the warmer ones. Data categorized according to province, reveal the highest mean annual OC
concentration in PD (6.8 µg m-3
), while the lowest mean value in TV (4.5 µg m-3
). OC
concentrations varied slightly among the provinces and differences were not statistically significant
as confirmed by a Kruskal–Wallis one-way analysis of variance (p-value 0.08). However, variations
for EC values among the provinces were statistically significant (Kruskal–Wallis one-way analysis
of variance test, p-value 0.00).
12
The combined effect of the lower air temperature favouring the particle-phase partitioning of semi-
volatile organics and the additional emission sources in the form of biomass burning or household
heating systems are possibly the main factors for higher OC concentrations in winter months (e.g.,
Viana et al., 2007; Schwarz et al., 2008). Moreover, the daily variability of OC concentrations is
higher in winter than in summer, probably because of significant local contributions with
remarkable daily variability, whereas homogenous contributions from regional-scale secondary
organic aerosols are the reason behind lower variability of OC concentrations in summer months
(Viana et al., 2007). The origin of carbonaceous aerosols can be evaluated from the relationship
between OC and EC (Turpin and Huntzicker, 1995). A strong relationship between OC and EC is
the indication of dominant primary sources (e.g. vehicle emissions, cooking, various combustion
processes). In this study, a statistically significant positive correlation (r=0.76) between OC and EC
was observed for all data in absence of any categorization. Before statistical analysis, normality of
the data was tested and data of OC and EC were not normally distributed. This was confirmed by
observing histograms, boxplots and finally by applying the Shapiro-Walk tests (assumption, p <
0.05 was not met). LogOC and logEC were used instead of OC and EC during statistical analysis as
the normality assumption was not met and correlation analysis was performed using a robust
procedure. The robust procedure description is provided as Supplementary Information (SI1).
By separating the data according to months, positive correlations between OC and EC were
observed in all the months except December (Table 4). However, the correlation is comparatively
higher in the warmer months than in the colder ones. This weaker correlation and higher OC/EC
ratio during the winter months is probably due to the influence of SOC (Secondary Organic Carbon)
formation or other significant primary organic sources (Na et al., 2004). This result is similar to the
one obtained by Kim et al. (2000), who suggested that the OC and EC may not be emitted from a
single dominant primary source (Na et al., 2004). The relationship between organic and elemental
carbon for all provinces was also determined (Table 4).
13
An effort was made to compare the levels of carbonaceous particulate matter in weekdays and
weekend (Figure SI2.1). Generally weekdays showed statistically significant higher EC and TC
concentrations as compared to weekend, as confirmed by Mann-Whitney U test (p value = 0.000
for EC; p value = 0.02 for TC), whereas the differences between weekdays and weekend were not
statistically significant for OC (p value = 0.069). All the measurement sites showed the same
pattern: the highest TC concentration was on Wednesday, whereas the lowest one was observed on
Sunday. This indicates nonappearance of pollution from commuting and other professional
activities (Martellini et al., 2012). As the normality assumption (p>0.05) was not met by Shapiro-
Wilk test, a non-parametric Kruskal-Wallis one way analysis of variance was performed to test the
significance of variations among days. Finally, a pair-wise comparison was carried out using the
Wilcoxon rank sum test with Bonferroni correction and the results show that the TC levels of
Wednesday are significantly different from those of Sunday (p value, 0.045). The highest OC
concentrations were on Wednesday, whereas the highest EC concentrations on Thursday. Moreover,
both OC and EC concentrations were higher on Saturday than on Monday, probably because most
of the business operations are continuing on Saturday in Italy.
3.4 Factors affecting organic and elemental carbon levels
A correlation analysis between some weather parameters with organic and elemental carbon data
was performed following a robust procedure in order to identify the factors affecting carbonaceous
compounds variations (Table 5). OC and EC had a statistically significant negative relationship with
temperature and solar radiation. This result suggests an increased emission during the winter
months and it is similar to the findings of Li et al. (2012). OC and EC also showed a negative
relationship with wind velocity. This is probably due to the clearing function of wind (Pindado et
al., 2009). Similar results were also reported for PAHs measured across the Veneto region (Masiol
14
et al., 2013). A weak positive correlation was observed for both OC and EC concentrations with
relative humidity. The absorption of pollutants may increase with the rises of humidity (Elbayoumi
et al., 2013). For illustration purpose, data have been categorized into two groups to be consistent
with the dates in which domestic heating is switched off (15 April) and on (15 October) according
to the national legislation: warm period (from April to October 14) and cold period (from October
15 to March). The correlation of OC with temperature was positive in the warm period, negative in
the cold one. A positive correlation between OC with temperature during the warm period may be
due to the contribution of the secondary organic carbon (Grivas et al., 2004), whereas a negative
correlation between OC and temperature may be related to the increased household heating during
the cold period. An additional explanation for this negative correlation can also be found in the
presence of lower pollutant dispersions with stable atmospheric condition during winter
(Vardoulakis and Kassomenos, 2008). A negative correlation between wind velocity and organic
and elemental carbon was observed in both the seasons.
Finally, statistically significant weather factors controlling OC and EC were investigated by fitting
linear models using a robust procedure based on weighted likelihood (Markatou et al., 1998).
Robust techniques enable to overcome small deviations from normality and the presence of few
outliers. A robust model selection procedure based on a weighted Akaike Information Criterion
(Agostinelli, 2001) was also used, to choose the best model. This turned out to be the model
including only temperature and wind. An additional confirmation was also provided by the robust t-
test based on the full model in which temperature, humidity, wind velocity and solar radiation were
considered as meteorological variables.
The selected model expression is the following:
15
)()log()log(6
1
321 i
i
i MonthIWindeTemperaturECOC …….. (1)
A robust R2 of the selected model is 0.79. The p-value and level of significance of the model
parameters and other related information is provided as Supplementary Information (Table SI2.1,
Figure SI2.2 and SI2.3). This procedure identifies three outliers, namely PD 39, 40 and VE 39 (see,
SI2). After checking those observations, we found that comparatively higher relative humidity
values were observed in these days and sampling sites (in PD, RH>91% and VE, RH>95%). This
may be one of the reasons.
3.5 Secondary organic Aerosol estimation
The quantification of primary and secondary organic carbon is quite difficult since no direct
analytical methods are available. Several indirect methods have been used to estimate the secondary
organic carbon. Among them, the EC tracer method is a widely accepted one: EC is used as a
tracer of Primary Organic Carbon (POC) (Turpin and Huntzicker, 1995; Cabada et al., 2004). SOC
is calculated by the following equation:
SOC=OCtot-EC (OC/EC)pri …………..(2)
Where OCsec is the secondary OC and OCtot is the measured total OC. Primary organic carbon
(POC) is calculated from the EC (OC/EC)pri. However, without conducting a complete profile of the
organic carbon and emission source, it is difficult to define the contribution of the primary source.
Several authors have recommended to substitute the minimum OC/EC ratio for the primary OC/EC
(Castro et al., 1999):
SOC=OCtot-EC (OC/EC)min ………….(3)
16
Based on this approach, the minimum OC/EC ratio of this study is 1.29, which is the mean value of
the lowest 5% OC/EC value. The mean annual primary and secondary organic carbon
concentrations are 1.7 µg m-3
and 3.8 µg m-3
, respectively (Table 6). In the cold period
comparatively higher secondary organic carbon values were found than in the warm one. For
illustration purpose, all data have been divided into two periods; warm period (April to October 14)
and cold period (October 15 to March). The cold period mean SOC concentration is 6.8 µg m-3
,
contributing for 76% of the cold period TOC (Total Organic Carbon), while the warm period SOC
(1.4 µg m-3
), contributes, on average, for 52% to the warm period TOC.
The EC/OC minimum ratio method and in general the EC tracer method have a limitation in the fact
that they do not account for the variability of primary sources and do not consider non-combustion
OC associated with EC, which is mainly of biogenic origin. A regression approach has been
suggested to overcome this limitation (Turpin and Huntzicker, 1995):
SOC=OC-POC=OC-[aEC+b]
where a is the primary combustion ratio and b is the non-combustion primary OC. Basing on this
approach, the annual mean of the SOC is 3.8 µg m-3
, contributing for 69% of the total organic
carbon; the warm period SOC is 1.3 µg m-3
, contributing for 49% of the warm period OC and the
cold period SOC is 6.4 µg m-3
, contributing for 72% of the cold period OC (Table 6).
It is interesting to observe that SOC is higher in the cold period than in the warm one, differently
from the expected pattern of higher concentration in the warm period as a consequence of
favourable meteorological conditions and the occurrence of photochemical reactions. However, the
seasonal variation of SOC follows the same trend of total OC with higher concentrations during the
17
household heating season, when higher emission of organic precursors occur. Moreover, a lower
mixing layer height in winter favours SOC precursor stagnation and SOC formation (Dan et al.,
2004; Duan et al., 2005). Under this view, a recent study carried out in VE by Pecorari et al. (2013)
showed that the lower mixing height during winter and autumn tend to confine pollutants within
500 m of height. In summer, the mixing height reaches 1 km or more and therefore allows a greater
dispersion of pollutants.
3.6 Conditional probability function (CPF) and conditional bivariate probability function
(CBPF)
Among the six measurement sites, BL and VE have been selected to investigate the potential local
sources of carbonaceous PM. BL is an interesting case of study for a number of reasons: (i) it is
located in an alpine valley surrounded by high mountains, mostly peaking at ~800-2000 m; (ii) the
atmospheric circulation is strongly affected by the arrangement of mountains; (iii) the sampling site
is located in a public park in the middle of the city; (iv) both hardwood (from local trees) and
softwood (mostly from conifers) are widely burned for domestic heating in stoves, most of which
are dated, i.e. built with old technologies and without any system for emission mitigation; (v) road
traffic is moderate; (vi) the site is located ~500 m East from a railway station, where trains are all
diesel-powered; (vii) there are no large industries. On the contrary, VE has opposed characteristics:
(i) it is located in a completely flat area; (ii) the wind regimes are affected by sea/land breezes in the
warm seasons; (iii) the sampling site is located in the NE part of the city; (iv) methane is commonly
used for domestic heating, but there has been a recent increase in the use of wood (i.e. logs,
briquettes, chips and pellets) in modern stoves; (v) road traffic is intense with frequent congestion
of the main roads in rush hours; (vi) a railway station is located ~2.6 km SSE (mostly electric-
powered trains) and an international airport at ~6 km E; (vii) a large industrial area is hosted in the
southern part of the city, including a main thermoelectrical power plant, oil refineries, incineration
18
facilities, chemical and steel plants. City maps with highlighted sources are reported as Figure
SI2.4.
CPF plots for EC, OC concentrations and OC/EC ratios in BL and VE are provided in Figure SI2.5,
whereas CBPF plots are shown in Figure 3. The plots for BL emphasize sectors, where the
concentration is >75th
percentile, equal to 1.5 and 6 µg m-3
for EC and OC, respectively. While CPF
evidences that higher probabilities of high concentration may originate from all the directions,
CBPF plots clearly show that the higher concentrations of EC and OC occur under wind calm
conditions and for very low winds blowing from S (toward the historic city center). This wind
regime corresponds to the prevailing wind direction in colder months, especially in December and
February (Figure SI2.6) and indicates that domestic heating is the main source of both OC and EC.
This result is also confirmed by analyzing the plot for OC/EC ratio. Both emission studies (e.g.,
Hildemann et al., 1991; Watson et al., 2001; Schauer et al., 2001; 2002) and monitoring campaigns
at urban and rural sites (e.g, Schwartz et al., 2008; Harrison and Yin, 2008; de la Campa et al.,
2009; Pirovano et al., 2015) reported larger proportion of OC emitted by biomass burning and
domestic heating with respect to mobile sources. In particular, OC/EC ratios ranging from 1 to 4
have been reported for diesel- and gasoline-powered vehicles (Schauer et al., 2002), while OC/EC
varying from 17 to 40 for wood combustion (Schauer et al., 2001). In BL, high probabilities of
OC/EC >6.4 are evidenced toward the city centre, confirming the relative dominance of OC derived
from biomass and wood burning. In the opposite way, despite mobile sources seem not to be the
main sources of carbonaceous PM2.5, their role as strong EC sources cannot be disregarded: the
CBPF clearly shows that lower probabilities of high OC/EC ratios are measured to NE, i.e. toward a
main traffic road often congested during weekends (most malls, shops and stores are to NE) and to
WNW, i.e. toward main roads and the railway (diesel-powered trains only) and bus station.
19
CPF plots for VE indicate high probability of both EC and OC >75th percentile (i.e., 1.3 and 9.7 µg
m-3
, respectively) for winds blowing from WSW and NW, while CBPF plots (Figure 3) show high
probabilities mainly for winds blowing from WSW, i.e. toward the main urban area of Mestre,
which also hosts some main roads affected by heavy traffic and an orbital road (Figure SI2.4). Some
insights seem to point out that vehicular traffic is the main source of carbonaceous PM2.5 in VE:
The results are consistent with NO2 data collected during 2000-2013 in the same site
(Masiol et al., 2014). NO2 is largely emitted by road traffic in urban environments and a
recent increase in NO2 levels in Europe has been related to the growing proportion of diesel-
powered vehicles, which are known to have higher primary (direct) emissions of NO2
(Carslaw et al., 2007);
CBPF clearly shows that probabilities of high OC/EC ratios (>6.3) are measured mainly
toward W and also spreading over the 1st and 2nd quadrant, i.e. toward the suburbs and
semirural areas surrounding the city, where the use of alternative fuels to methane for
domestic heating is increasing.
Finally it could be concluded that both biomass & wood burning and vehicular traffic are probably
the main local sources of carbonaceous particulate matter in Veneto region.
3.7 Potential effects of long-range transports
In this study, four back-trajectories were simulated for each day and site in the flat areas of the
region (VI, VE, PD and RO) as the topography of the territory may have effects on the
reconstructions of back-trajectories for BL and TV, which are in and close to the Alps, respectively.
The optimum number of clusters was then established by analyzing the change in the total spatial
variance: all sites exhibited five main clusters with similar origins. Clusters (Figure 4) were then
identified according to their provenance as: (1) Western Europe, (2) Mediterranean, (3) local, (4)
Northern Europe and (5) Eastern Europe. The daily OC, EC and TC concentrations have been
20
subsequently assigned to each of the clustered trajectories. Statistics for chemical composition data
in each cluster are presented in Table 7.
Generally, OC, EC and TC show similar results at all the sites, with concentrations slightly higher
for clusters 1, 4 and 5 and the inter-cluster comparison shows only small differences. In particular,
the site RO can be roughly selected as the most reliable one for analysing the long-range transports,
as it is located far away from the mountain chains and the city has fewer local industrial emissions
and traffic. In RO, the highest levels of OC are associated with air masses blowing from Northern
(6.7 μg m-3
) and Eastern (5.7 μg m-3
) Europe, while results for the remaining origins are quite
constant (4-5 μg m‒3
). In summary, results indicate that trans-boundary transports have only a
limited influence on the levels of OC and EC in the Veneto region.
4. CONCLUSIONS
In this work, for the first time, OC and EC concentrations were investigated in PM2.5 in the Veneto
region for an extended period of time. The mean organic carbon concentration was 5.48 µg m-3
,
contributing for almost 22% of the total PM2.5 and 79% of the total carbon, while the mean EC
value was 1.31 µg m-3
, contributing for 5% of the total PM2.5 and 21% to the total carbon. The
monthly OC concentration gradually increased from April to December. The EC did not vary in
accordance with OC, but the highest values were recorded in the colder months, as well. The
OC/EC ratios ranged from 0.71 to 15.38 with a mean value of 4.54, which is higher than the values
observed in most of the other European cities. In this study, a positive statistically significant
relationship (r=0.76) between elemental and organic carbon was observed without any
categorization of the data. However, the correlation is comparatively better in the warmer months
than in the colder ones. A weak correlation coefficient between elemental and organic carbon
during colder months suggests that the OC and EC may not be emitted from a single dominant
primary source. Statistically significant micro-meteorological factors controlling organic and
21
elemental carbon were investigated by fitting linear models using a robust procedure based on
weighted likelihood. Temperature and wind velocity turned out to be significantly related to the
carbon fractions with a multiple R2 value of 0.79. The secondary organic carbon contributed for
69% of the total organic carbon during the study period as confirmed by both the approaches of
OC/EC minimum ratio and regression. It is interesting to observe that SOC is higher in the cold
period than the warm one (in contrast with the expected pattern of higher concentration in summer),
as a consequence of lower temperature and stable atmospheric conditions favouring accumulation
of pollutants and hastening the condensation or nucleation of volatile organic compounds. The local
contribution to OC and EC was investigated by using both CPF and CBPF. Plots indicate that both
biomass & wood burning and vehicular emissions are probably the main local sources of
carbonaceous particulate matter. The potential effects of long-range transports on the OC and EC
levels were analysed by clustering the back-trajectories in the sites located in the flat areas of the
region. Results revealed no significant differences in the levels of both OC and EC, when air masses
had passed across different European regions, indicating that trans-boundary transports have a
limited effect on the carbonaceous PM variations in the Veneto region.
ACKNOWLEDGEMENTS
This study was the result of an agreement between the Ca’ Foscari University of Venice and the
Regional Agency for Environmental Protection of Veneto (ARPAV). The authors are grateful to the
NOAA Air Resources Laboratory (ARL) for providing the HYSPLIT transport and dispersion
model and/or the READY website (http://www.ready.noaa.gov) used in this publication. Finally,
Dr. M. Masiol takes the opportunity to express his gratefulness to the European Union for funding
the Marie Curie Intra-European Fellowships (Project CHEERS, Cal: PEOPLE-2012-IEF).
DISCLAIMER
22
This study was not financially supported by any public or private institution. The authors wish to
state that the views expressed in this study are exclusively of their own and may not necessarily
coincide with those of ARPAV.
REFERENCES
Agostinelli C. Test of hypotheses based on the weighted likelihood methodology. Stat Sinica 2001;
11:499-51.
Ashbaugh II, Malm WC, Sadeh WZ. A residence time probability analysis of sulfur concentrations
at grand Canyon National Park. Atmos Environ 1985; 19 (8):1263-70.
Begum BA, Biswas SK, Markwitz A, Hopke PK. Identification of sources of fine and course
particulate matter in Dhaka, Bangladesh. Aerosol Air Qual Res 2010; 10:345-53.
Bigi A, Ghermandi G, Harrison RM. Analysis of the air pollution climate at a background site in
the Po valley. J Environ Monit 2012 14; 552-563.
Birch ME, Cary RA. Elemental carbon-based method for monitoring occupational exposures to
particulate diesel exhaust. Aerosol Sci Tech 1996; 25:221-41.
Cabada J, Pandis SN, Subramanian R, Robinson AL, Polidori A, Turpin B. Estimating the
secondary organic aerosol contribution to PM2.5 using the EC tracer method. Aerosol Sci Tech
2004; 38:140–55.
Carslaw DC. The openair manual — open-source tools for analysing air pollution data. Version
10th June 2014, King’s College London.
Carslaw DC, Beevers SD, Bell MC. Risks of exceeding the hourly EU limit value for nitrogen
dioxide resulting from increased road transport emissions of primary nitrogen dioxide. Atmos
Environ 2007; 41: 2073-82.
Castro LM, Pio CA, Harrison RM, Smith DJT. Carbonaceous aerosol in urban and rural European
atmospheres: estimation of secondary organic carbon concentrations. Atmos Environ 1991; 33:
2771–81.
Cuccia E, Massaò D, Ariola V, Bove MC, Fermo P, Piazzalunga A, Prati P. Size-resolved
comprehensive characterization of airborne particulate matter. Atmos Environ 2013; 67: 14-26.
Dan M, Zhuang G, Li X, Tao H, Zhuang Y. The characteristics of carbonaceous species and their
sources in PM2.5 in Beijing. Atmos Environ 2004; 38(21):3443–52.
23
de la Campa AS, Pio C, de La Rosa JD, Querol X, Alastuey A, González-Castanedo Y.
Characterization and origin of EC and OC particulate matter near the Doñana National Park (SW
Spain). Environ Res 2009; 109(6):671-681.
Di Giuseppe F, Riccio A, Caporaso L, Bonafe G, Gobbi GP, Angelini F. Automatic detection of
atmospheric boundary layer height using ceilometers backscatter data assisted by a boundary layer
model. Q J R. Meteorol Soc 2012; 138:649-663.
Draxler RR, Rolph GD. HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory)
Model access via NOAA ARL READY Website (http://www.arl.noaa.gov/HYSPLIT.php). NOAA
Air Resources Laboratory, College Park, MD; 2013.
Duan F, He K, Ma Y, Jia Y, Yang F, Lei Y et al. Characteristics of carbonaceous aerosols in
Beijing, China. Chemosphere 2005; 60:355–64.
Elbayoumi M, Ramli NA, Yusuf NFFM, Madhoun WA. Spatial and seasonal variation of
particulate matter (PM10 and PM2.5) in Middle Eastern classrooms. Atmos Environ 2013; 80:389-
397.
Ferrero L, Perrone MG, Petraccone S, Sangiorgi G, Ferrini BS, Lo Porto C et al. Vertically resolved
particle size distribution within and above the mixing layer over the Milan metropolitan area.
Atmos Chem Phys 2010; 10:3915–32.
Giannoni M, Martellini T, Bubba MD, Gambaro A, Zangrando R, Chiari M et al. The use of
levoglucosan for tracing biomass burning in PM2.5 samples in Tuscany (Italy). Environ Pollut
2012; 167:7-15.
Gilardoni S, Vignati E, Cavalli F, Putaud JP, Larsen BR, Karl M et al. Better constraints on sources
of carbonaceous aerosols using a combined C-14-macro tracer analysis in a European rural
background site. Atmos Chem Phys 2011; 11:5685-00.
Grivas G, Chaloulakou A, Samara C, Spyrellis N. Spatial and temporal variation of PM10 mass
concentrations within the greater area of Athens, Greece. Water Air Soil Pollut 2004; 158:357–71.
Harrison RM, Yin J. Sources and processes affecting carbonaceous aerosol in central England.
Atmos Environ 2008; 42(7):1413–23.
Hildemann LM., Markowski GR, Cass GR. Chemical composition of emissions from urban sources
of fine organic aerosol. Environ Sci Technol 1991; 25:744–759.
Hitzenberger R, Jennings SG, Larson SM, Dillner A, Cachier H, Galambos Z et al. Intercomparison
of measurement methods for black carbon aerosols. Atmos Environ 1999; 33:2823–33.
Ito K, Mathes R, Ross Z, Nadas A, Thurston G, Matte T. Fine particulate matter constituents
associated with cardiovascular hospitalizations and mortality in New York City. Environ Health
Perspect 2011; 119:467–73.
Jacobson MC, Hansson HC, Noone KJ, Charlson RJ. Organic atmospheric aerosols: Review and
state of the science. Rev Geophys 2000; 38:267–94.
24
Jedynska A, Hoek G, Eeftens M, Cyrys J, Keuken M, Ampe et al. Spatial variations of PAH,
hopanes/steranes and EC/OC concentrations within and between European study areas. Atmos
Environ 2014; 87:239-248.
Kim BM, Teffera S, Zeldin MD. Characterization of PM2.5 and PM10 in the south coast air basin of
Southern California: Part 1—spatial variations. J Air Waste Manag Assoc 2000; 50:2034–44.
Kim E, Hopke PK, Edgerton ES. Source identification of Atlanta aerosol by positive matrix
factorization. J Air Waste Manag Assoc 2003; 53:731-39.
Kim E, Hopke PK, Kenski DM, Koerber M. Sources of fine particles in a rural Midwestern U.S.
area. Environ Sci Technol 2005; 39:4953-60.
Larsen BR, Gilardoni S, Ktenstrom K, Niedzialek J, Jimenez J, Belis CA. Sources for PM air
pollution in the Po Plain, Italy: II. Probabilistic uncertainty characterization and sensitivity analysis
of secondary and primary sources. Atmos Environ 2012; 50:203-213.
Li P, Han B, Huo J, Lu B, Ding X, Chen L et al. Wang, B. Characterization, meteorological
influences and source identification of carbonaceous aerosols during the autumn-winter period in
Tianjin, China. Aerosol Air Qual Res 2012; 12:283-94.
Lim H, Turpin B. Origins of primary and secondary organic aerosol in Atlanta: results of time-
resolved measurements during the Atlanta Supersite experiment. Environ Sci Technol 2002;
36:4489–96.
Lonati G, Giugliano M, Butelli P, Romele L, Tardivo R. Major chemical components of PM2.5 in
Milan (Italy). Atmos Environ 2005; 39:1925–34.
Lonati G, Ozgen S, Giugliano M. Primary and secondary carbonaceous species in PM2.5 samples in
Milan (Italy). Atmos Environ 2007; 41 (22): 4599–10.
Lyamani H, Olmo FJ, Alcàntara A, Alados-Arboledas L. Atmospheric aerosols during the 2003
heat wave in southeastern Spain II: microphysical columnar properties and radiative forcing. Atmos
Environ 2006; 40:6465–76.
Markatou M, Basu A, Lindsay BG. Weighted likelihood estimating equations with a bootstrap root
search. J Am Stat Assoc 1998; 93:740-50.
Malaguti A, Mircea M, Torretta TMGL, Piersanti A, Salvi S, Zanini G et al. Fine carbonaceous
aerosol characteristics at a coastal rural site in the Central Meditterranean as given by OCEC online
measurement. Journal of Aerosol Sciences 2013; 56:78-87.
Martellini T, Giannoni M, Lepri L, Katsoyiannis A, Cincinelli A. One year intensive PM2.5 bound
polycyclic aromatic hydrocarbons monitoring in the area of Tuscany, Italy. Environ Pollut 2012;
164:252-58.
Masiol M, Formenton G, Pasqualetto A, Pavoni B. Seasonal trends and spatial variations of PM10-
bounded polycyclic aromatic hydrocarbons in Veneto region, Northeast Italy. Atmos Environ 2013;
79:811-821.
25
Masiol M, Agostinelli C, Formenton G, Tarabotti E, Pavoni B. Thirteen years of air pollution
hourly monitoring in a large city: Potential sources, trends, cycles and effects of car-free days. Sci
Tot Environ 2014; 494–495:84–96.
Na K, Sawant AA, Song C, Cocker DR. Primary and secondary carbonaceous species in the
atmosphere of Western Riverside County, California. Atmos Environ 2004; 38 (9):1345–55.
Pecorari E, Squizzato S,Masiol M, Radice P, Pavoni B, Rampazzo G. Using a photochemical model
to assess the horizontal, vertical and time distribution of PM2.5 in a complex area: relationships
between the regional and local sources and the meteorological conditions. Sci Total Environ 2013;
443:681–91.
Perrone MR, Piazzalunga A, Prato M, Carofalo I. Composition of fine and coarse particles in a
coastal site of the central Mediterranean: Carbonaceous species contributions. Atmos Environ 2011;
45:7470-77.
Pindado O, Perez R, Garcia S, Sanchez M. Galan P, Fernandez M. Characterization and Sources
Assignation of PM2.5 Organic Aerosol in a Rural Area of Spain. Atmos Environ 2009; 43:2796–03.
Pirovano G, Colombi C, Balzarini A, Riva GM, Gianelle V, Lonati G. PM2.5 source apportionment
in Lombardy (Italy): Comparison of receptor and chemistry-transport modelling results. Atmos
Environ 2015; 106:56-70.
Putaud JP, Raes F, van Dingenen R, Bruggemann E, Facchini MC, Decesari S, et al. A European
aerosol phenomenology — 2: Chemical characteristics of particulate matter at kerbside, urban, rural
and background sites in Europe. Atmos Environ 2004; 38:2579–95.
Rolph GD. Real-time Environmental Applications and Display System (READY) Website
(http://www.ready.noaa.gov). NOAA Air Resources Laboratory, College Park, MD; 2013.
Salma I, Chi X, Maenhaut, W. Elemental and organic carbon in urban canyon and background
environments in Budapest, Hungary. Atmos Environ 2004; 38:27–36.
Sandrini N, Fuzzi S, Piazzalunga A, Prati P, Bonasoni P. Cavalli et al. Spatial and seasonal
variability of carbonaceous aerosol across Italy. Atmos Environ 2014; 99:587-98.
Schauer JJ, Kleeman MJ, Cass GR, Simoneit BRT. Measurement of emissions from air pollution
sources. 3. C1–C29 organic compounds from fireplace combustion of wood. Environ. Sci. Technol
2001; 35:1716–28.
Schauer JJ, Kleeman MJ, Cass GR, Simoneit BRT. Measurement of emissions from air pollution
sources. 5. C1–C32 organic compounds from gasoline-powered motor vehicles. Environ. Sci. Tech
2002; 36:1169–80.
Schwarz J, Chi X, Maenhaut W, Civis M, Hororka J, Smolik J. Elemental and organic carbon in
atmospheric aerosols at downtown and suburban sites in Prague. Atmos Res 2008; 90:287-02.
Seinfeld JH, Pandis SN. In: Atmospheric Chemistry and Physics: From Air Pollution to Climate
Change. New York: Wiley; 1998.
26
Turpin BJ, Huntzicker JJ. Identification of secondary organic aerosol episodes and quantification of
primary and secondary organic aerosol concentrations during SCAQS. Atmos Environ 1995;
29:3527–44.
Uria-Tellaetxe I, Carslaw DC. Conditional bivariate probability function for source identification.
Environ Model Softw 59 2014; 59:1-9.
Vardoulakis S, Kassomenos P. Sources and factors affecting PM10 levels in two European cities:
Implications for local air quality management. Atmos Environ 2008; 42: 3949-63.
Viana M, Maenhaut W, Brink HM, Chi X, Weijers E, Querol X et al.Comparative analysis of
organic and elemental carbon concentrations in carbonaceous aerosols in three European cities.
Atmos Environ 2007; 41:5972–83.
Viidanoja J, Sillanpaa M, Laakia J, Kerminen V, Hillamo R, Aarnio P, Koskentalo T. Organic and
black carbon in PM2.5 and PM10: 1 year of data from an urban site in Helsinki, Finland. Atmos
Environ 2002; 36:3183–3193.
Watson JG, Chow JC, Houck JE. PM2.5 chemical source profiles for vehicle exhaust, vegetative
burning, geological material, and coal burning in northwestern Colorado during 1995, Chemosphere
2001; 43(8):1141–51.
WHO (World Health Organization). Air quality guidelines for Europe. 2nd ed. Copenhagen: WHO
Regional Publications; 2000. European Series No. 91.